원클릭으로 Manus에서 모든 스킬 실행

debug-verify-benchmark

스타4

포크0

업데이트2026년 1월 31일 04:37

Debug, verify, and compare elix-db to industry after each plan step. Use after implementing any plan step or changing vector/store/API logic; run tests, IEx checks, and document efficiency vs Qdrant/Milvus/pgvector.

설치

Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.

Manus에서 실행

출처

8dazo

8dazo/elix-db

GitHub 저장소 열기 Creator 저장소 보기

다운로드

Manus에서 실행

Debug, Verify, and Benchmark

Apply after every plan step or change to vector store, search, or API.

When to Use

After implementing a step in plan/README.md
After changing ElixDb vector/store/API logic
Before marking a step "Done" in the plan

Debug

Run tests: mix test
Start IEx: iex -S mix and exercise the new APIs (create collection, upsert, search, get, delete as applicable).
Add traces if needed: Logger.debug/2 or :sys.trace for GenServer; fix failures before proceeding.

Verify

All tests pass; no new compiler warnings.
For search: add or run tests that check known vectors produce known ordering (e.g. cosine similarity order).
Acceptance criteria in the step file are checked off.

Industry Comparison

For each step, fill in the Industry comparison table in the step file:

Correctness: Behavior matches Qdrant/Milvus (collections, points, upsert, search, get, delete).
Latency: Note expected vs actual; compare to typical Qdrant/Milvus/pgvector numbers (e.g. p99 ms) when metrics exist.
Throughput: QPS if measured; document concurrency level.
Recall: For search, recall@k = 1.0 for exact k-NN; document when approximate index is added later.

Document efficiency notes at the bottom of the step: gaps, improvement ideas, and follow-up tasks.

Metrics to Capture

When implementing step 8 or ad-hoc benchmarking:

Metric	How	Use
Latency	Per-operation timing; compute mean, p50, p99	Compare to industry; track over time
QPS	Queries per second under fixed concurrency	Throughput vs Qdrant/Milvus
Recall@k	True k-NN vs returned k; fraction overlap	Search quality when ground truth exists
Memory/CPU	BEAM process stats	Resource efficiency

Store results in a simple format (e.g. JSON struct or Markdown table) under docs/benchmarks.md or script output for future comparison.

이 저장소의 다른 Skills

같은 저장소

versioned-sample-uses

8dazo/elix-db

When publishing a new elix-db version or making version changes, run versioned sample use cases, collect memory/time/latency, compare to the previous version report, and update reports. Remember this workflow for releases.

2026-01-314

elixir-vector-patterns

8dazo/elix-db

Elixir patterns for vector storage and similarity search. Use when implementing GenServer-based vector store, ETS index, Nx/Scholar distance math, or supervision for a vector DB in Elixir.

2026-01-314

vector-db-elixir

8dazo/elix-db

Design and implement a vector database in Elixir. Use when building embedding storage, similarity search, k-NN retrieval, or when the user mentions vector DB, embeddings, or semantic search in Elixir.

2026-01-314

vector-db-storage

8dazo/elix-db

Storage backends and indexing for vector databases in Elixir. Use when choosing persistence, implementing exact vs approximate k-NN, or integrating pgvector, file, or ETS for vector storage.

2026-01-314

name	debug-verify-benchmark
description	Debug, verify, and compare elix-db to industry after each plan step. Use after implementing any plan step or changing vector/store/API logic; run tests, IEx checks, and document efficiency vs Qdrant/Milvus/pgvector.

Debug, Verify, and Benchmark

Apply after every plan step or change to vector store, search, or API.

When to Use

After implementing a step in plan/README.md
After changing ElixDb vector/store/API logic
Before marking a step "Done" in the plan

Debug

Run tests: mix test
Start IEx: iex -S mix and exercise the new APIs (create collection, upsert, search, get, delete as applicable).
Add traces if needed: Logger.debug/2 or :sys.trace for GenServer; fix failures before proceeding.

Verify

All tests pass; no new compiler warnings.
For search: add or run tests that check known vectors produce known ordering (e.g. cosine similarity order).
Acceptance criteria in the step file are checked off.

Industry Comparison

For each step, fill in the Industry comparison table in the step file:

Correctness: Behavior matches Qdrant/Milvus (collections, points, upsert, search, get, delete).
Latency: Note expected vs actual; compare to typical Qdrant/Milvus/pgvector numbers (e.g. p99 ms) when metrics exist.
Throughput: QPS if measured; document concurrency level.
Recall: For search, recall@k = 1.0 for exact k-NN; document when approximate index is added later.

Document efficiency notes at the bottom of the step: gaps, improvement ideas, and follow-up tasks.

Metrics to Capture

When implementing step 8 or ad-hoc benchmarking:

Metric	How	Use
Latency	Per-operation timing; compute mean, p50, p99	Compare to industry; track over time
QPS	Queries per second under fixed concurrency	Throughput vs Qdrant/Milvus
Recall@k	True k-NN vs returned k; fraction overlap	Search quality when ground truth exists
Memory/CPU	BEAM process stats	Resource efficiency

Store results in a simple format (e.g. JSON struct or Markdown table) under docs/benchmarks.md or script output for future comparison.